File size: 13,864 Bytes
7fcd557 24a059f c308901 7fcd557 c308901 24a059f 7fcd557 24a059f c308901 24a059f 7fcd557 24a059f 7fcd557 24a059f 7fcd557 24a059f 7fcd557 e4f66e8 24a059f e4f66e8 7fcd557 24a059f e4f66e8 7fcd557 24a059f c308901 7fcd557 24a059f 7fcd557 e4f66e8 c308901 ca6bd07 9f7748a ca6bd07 9f7748a ca6bd07 9f7748a ca6bd07 9f7748a 8f89713 104bf5a 8f89713 104bf5a 8f89713 7fcd557 c308901 8f89713 7fcd557 8f89713 7fcd557 104bf5a ca6bd07 8f89713 ca6bd07 8f89713 7fcd557 8f89713 7fcd557 104bf5a 7fcd557 8f89713 7fcd557 8f89713 7fcd557 8f89713 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 |
# import gradio as gr
# import pandas as pd
# import os
# import re
# from datetime import datetime
# LEADERBOARD_FILE = "leaderboard.csv" # File to store leaderboard data
# def clean_answer(answer):
# if pd.isna(answer):
# return None
# answer = str(answer)
# clean = re.sub(r'[^A-Da-d]', '', answer)
# if clean:
# first_letter = clean[0].upper()
# if first_letter in ['A', 'B', 'C', 'D']:
# return first_letter
# return None
# def write_evaluation_results(results, output_file):
# os.makedirs(os.path.dirname(output_file) if os.path.dirname(output_file) else '.', exist_ok=True)
# timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
# output_text = [
# f"Evaluation Results for Model: {results['model_name']}",
# f"Timestamp: {timestamp}",
# "-" * 50,
# f"Overall Accuracy (including invalid): {results['overall_accuracy']:.2%}",
# f"Accuracy (valid predictions only): {results['valid_accuracy']:.2%}",
# f"Total Questions: {results['total_questions']}",
# f"Valid Predictions: {results['valid_predictions']}",
# f"Invalid/Malformed Predictions: {results['invalid_predictions']}",
# f"Correct Predictions: {results['correct_predictions']}",
# "\nPerformance by Field:",
# "-" * 50
# ]
# for field, metrics in results['field_performance'].items():
# field_results = [
# f"\nField: {field}",
# f"Accuracy (including invalid): {metrics['accuracy']:.2%}",
# f"Accuracy (valid only): {metrics['valid_accuracy']:.2%}",
# f"Correct: {metrics['correct']}/{metrics['total']}",
# f"Invalid predictions: {metrics['invalid']}"
# ]
# output_text.extend(field_results)
# with open(output_file, 'w') as f:
# f.write('\n'.join(output_text))
# print('\n'.join(output_text))
# print(f"\nResults have been saved to: {output_file}")
# def update_leaderboard(results):
# # Add results to the leaderboard file
# new_entry = {
# "Model Name": results['model_name'],
# "Overall Accuracy": f"{results['overall_accuracy']:.2%}",
# "Valid Accuracy": f"{results['valid_accuracy']:.2%}",
# "Correct Predictions": results['correct_predictions'],
# "Total Questions": results['total_questions'],
# "Timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
# }
# leaderboard_df = pd.DataFrame([new_entry])
# if os.path.exists(LEADERBOARD_FILE):
# existing_df = pd.read_csv(LEADERBOARD_FILE)
# leaderboard_df = pd.concat([existing_df, leaderboard_df], ignore_index=True)
# leaderboard_df.to_csv(LEADERBOARD_FILE, index=False)
# def display_leaderboard():
# if not os.path.exists(LEADERBOARD_FILE):
# return "Leaderboard is empty."
# leaderboard_df = pd.read_csv(LEADERBOARD_FILE)
# return leaderboard_df.to_markdown(index=False)
# def evaluate_predictions(prediction_file):
# ground_truth_file = "ground_truth.csv" # Specify the path to the ground truth file
# if not prediction_file:
# return "Prediction file not uploaded", None
# if not os.path.exists(ground_truth_file):
# return "Ground truth file not found", None
# try:
# predictions_df = pd.read_csv(prediction_file.name)
# ground_truth_df = pd.read_csv(ground_truth_file)
# # Extract model name
# try:
# filename = os.path.basename(prediction_file.name)
# if "_" in filename and "." in filename:
# model_name = filename.split('_')[1].split('.')[0]
# else:
# model_name = "unknown_model"
# except IndexError:
# model_name = "unknown_model"
# # Merge dataframes
# merged_df = pd.merge(
# predictions_df,
# ground_truth_df,
# on='question_id',
# how='inner'
# )
# merged_df['pred_answer'] = merged_df['predicted_answer'].apply(clean_answer)
# invalid_predictions = merged_df['pred_answer'].isna().sum()
# valid_predictions = merged_df.dropna(subset=['pred_answer'])
# correct_predictions = (valid_predictions['pred_answer'] == valid_predictions['Answer']).sum()
# total_predictions = len(merged_df)
# total_valid_predictions = len(valid_predictions)
# overall_accuracy = correct_predictions / total_predictions if total_predictions > 0 else 0
# valid_accuracy = (
# correct_predictions / total_valid_predictions
# if total_valid_predictions > 0
# else 0
# )
# field_metrics = {}
# for field in merged_df['Field'].unique():
# field_data = merged_df[merged_df['Field'] == field]
# field_valid_data = field_data.dropna(subset=['pred_answer'])
# field_correct = (field_valid_data['pred_answer'] == field_valid_data['Answer']).sum()
# field_total = len(field_data)
# field_valid_total = len(field_valid_data)
# field_invalid = field_total - field_valid_total
# field_metrics[field] = {
# 'accuracy': field_correct / field_total if field_total > 0 else 0,
# 'valid_accuracy': field_correct / field_valid_total if field_valid_total > 0 else 0,
# 'correct': field_correct,
# 'total': field_total,
# 'invalid': field_invalid
# }
# results = {
# 'model_name': model_name,
# 'overall_accuracy': overall_accuracy,
# 'valid_accuracy': valid_accuracy,
# 'total_questions': total_predictions,
# 'valid_predictions': total_valid_predictions,
# 'invalid_predictions': invalid_predictions,
# 'correct_predictions': correct_predictions,
# 'field_performance': field_metrics
# }
# update_leaderboard(results)
# output_file = "evaluation_results.txt"
# write_evaluation_results(results, output_file)
# return "Evaluation completed successfully! Leaderboard updated.", output_file
# except Exception as e:
# return f"Error during evaluation: {str(e)}", None
# # Gradio Interface
# description = "Upload a prediction CSV file to evaluate predictions against the ground truth and update the leaderboard."
# demo = gr.Blocks()
# with demo:
# gr.Markdown("# Prediction Evaluation Tool with Leaderboard")
# with gr.Tab("Evaluate"):
# file_input = gr.File(label="Upload Prediction CSV")
# eval_status = gr.Textbox(label="Evaluation Status")
# eval_results_file = gr.File(label="Download Evaluation Results")
# eval_button = gr.Button("Evaluate")
# eval_button.click(
# evaluate_predictions, inputs=file_input, outputs=[eval_status, eval_results_file]
# )
# with gr.Tab("Leaderboard"):
# leaderboard_text = gr.Textbox(label="Leaderboard", interactive=False)
# refresh_button = gr.Button("Refresh Leaderboard")
# refresh_button.click(display_leaderboard, outputs=leaderboard_text)
# if __name__ == "__main__":
# demo.launch()
import gradio as gr
import pandas as pd
import os
import re
from datetime import datetime
LEADERBOARD_FILE = "leaderboard.csv" # File to store leaderboard data
LAST_UPDATED = datetime.now().strftime("%B %d, %Y")
def clean_answer(answer):
if pd.isna(answer):
return None
answer = str(answer)
clean = re.sub(r'[^A-Da-d]', '', answer)
if clean:
return clean[0].upper()
return None
def evaluate_predictions(prediction_file):
ground_truth_file = "ground_truth.csv"
if not os.path.exists(ground_truth_file):
return "Ground truth file not found."
if not prediction_file:
return "Prediction file not uploaded."
try:
predictions_df = pd.read_csv(prediction_file.name)
ground_truth_df = pd.read_csv(ground_truth_file)
model_name = os.path.basename(prediction_file.name).split('_')[1].split('.')[0]
merged_df = pd.merge(predictions_df, ground_truth_df, on='question_id', how='inner')
merged_df['pred_answer'] = merged_df['predicted_answer'].apply(clean_answer)
valid_predictions = merged_df.dropna(subset=['pred_answer'])
correct_predictions = (valid_predictions['pred_answer'] == valid_predictions['Answer']).sum()
total_predictions = len(merged_df)
total_valid_predictions = len(valid_predictions)
overall_accuracy = correct_predictions / total_predictions if total_predictions > 0 else 0
valid_accuracy = correct_predictions / total_valid_predictions if total_valid_predictions > 0 else 0
results = {
'model_name': model_name,
'overall_accuracy': overall_accuracy,
'valid_accuracy': valid_accuracy,
'correct_predictions': correct_predictions,
'total_questions': total_predictions,
}
update_leaderboard(results)
return "Evaluation completed successfully! Leaderboard updated."
except Exception as e:
return f"Error during evaluation: {str(e)}"
# Build Gradio App
def update_leaderboard(results):
"""
Update the leaderboard file with new results.
"""
new_entry = {
"Model Name": results['model_name'],
"Overall Accuracy": round(results['overall_accuracy'] * 100, 2),
"Valid Accuracy": round(results['valid_accuracy'] * 100, 2),
"Correct Predictions": results['correct_predictions'],
"Total Questions": results['total_questions'],
"Timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
}
# Convert new entry to DataFrame
new_entry_df = pd.DataFrame([new_entry])
# Append to leaderboard file
if not os.path.exists(LEADERBOARD_FILE):
# If file does not exist, create it with headers
new_entry_df.to_csv(LEADERBOARD_FILE, index=False)
else:
# Append without headers
new_entry_df.to_csv(LEADERBOARD_FILE, mode='a', index=False, header=False)
def load_leaderboard():
"""
Load the leaderboard from the leaderboard file.
"""
if not os.path.exists(LEADERBOARD_FILE):
return pd.DataFrame({
"Model Name": [],
"Overall Accuracy": [],
"Valid Accuracy": [],
"Correct Predictions": [],
"Total Questions": [],
"Timestamp": [],
})
return pd.read_csv(LEADERBOARD_FILE)
def evaluate_predictions_and_update_leaderboard(prediction_file):
"""
Evaluate predictions and update the leaderboard.
"""
ground_truth_file = "ground_truth.csv"
if not os.path.exists(ground_truth_file):
return "Ground truth file not found.", load_leaderboard()
if not prediction_file:
return "Prediction file not uploaded.", load_leaderboard()
try:
predictions_df = pd.read_csv(prediction_file.name)
ground_truth_df = pd.read_csv(ground_truth_file)
model_name = os.path.basename(prediction_file.name).split('_')[1].split('.')[0]
merged_df = pd.merge(predictions_df, ground_truth_df, on='question_id', how='inner')
merged_df['pred_answer'] = merged_df['predicted_answer'].apply(clean_answer)
valid_predictions = merged_df.dropna(subset=['pred_answer'])
correct_predictions = (valid_predictions['pred_answer'] == valid_predictions['Answer']).sum()
total_predictions = len(merged_df)
total_valid_predictions = len(valid_predictions)
overall_accuracy = correct_predictions / total_predictions if total_predictions > 0 else 0
valid_accuracy = correct_predictions / total_valid_predictions if total_valid_predictions > 0 else 0
results = {
'model_name': model_name,
'overall_accuracy': overall_accuracy,
'valid_accuracy': valid_accuracy,
'correct_predictions': correct_predictions,
'total_questions': total_predictions,
}
update_leaderboard(results)
return "Evaluation completed successfully! Leaderboard updated.", load_leaderboard()
except Exception as e:
return f"Error during evaluation: {str(e)}", load_leaderboard()
# Build Gradio App
with gr.Blocks() as demo:
gr.Markdown("# Prediction Evaluation Tool with Leaderboard")
with gr.Tabs():
# Submission Tab
with gr.TabItem("🏅 Submission"):
file_input = gr.File(label="Upload Prediction CSV")
eval_status = gr.Textbox(label="Evaluation Status", interactive=False)
leaderboard_table_preview = gr.Dataframe(
value=load_leaderboard(),
label="Leaderboard (Preview)",
interactive=False,
wrap=True,
)
eval_button = gr.Button("Evaluate and Update Leaderboard")
eval_button.click(
evaluate_predictions_and_update_leaderboard,
inputs=[file_input],
outputs=[eval_status, leaderboard_table_preview],
)
# Leaderboard Tab
with gr.TabItem("🏅 Leaderboard"):
leaderboard_table = gr.Dataframe(
value=load_leaderboard(),
label="Leaderboard",
interactive=False,
wrap=True,
)
refresh_button = gr.Button("Refresh Leaderboard")
refresh_button.click(
lambda: load_leaderboard(),
inputs=[],
outputs=[leaderboard_table],
)
gr.Markdown(f"Last updated on **{LAST_UPDATED}**")
demo.launch()
|